The question
The forecast is being asked something the pipeline cannot answer
Every CRM can tell you where a deal sits, what probability attaches to that stage, and what the weighted pipeline adds up to. None of that establishes that the opportunity is real. A deal can sit at Negotiation, 80%, closing this month, and be a conversation that ended six weeks ago.
The gap is not a reporting gap. Stage, probability and close date are all things a seller entered, on the seller’s schedule and with the seller’s incentives. Asking the forecast to grade its own inputs is asking the record to audit itself.
So the useful question is not "what does the pipeline say". It is "what did the buyer actually do, and does the pipeline agree with it".
Credit where it is due
What your CRM already does well, and where it stops
This is worth saying plainly, because the answer to a bad forecast is usually not another system. A modern CRM — HubSpot Sales Hub, Zoho, Salesforce — genuinely covers most of the forecasting job, and if your implementation is not using it, adding Klosure will not save you.
| Question | The CRM answers it | What is still missing |
|---|---|---|
| Where is the deal in the process? | Yes — stages, probability, weighted pipeline | Nothing. This is the CRM’s job and it does it. |
| Which deals are in Commit, Best case or Pipeline? | Yes — forecast categories, set by rule or by hand | Whether the category survives contact with the evidence |
| Which deals look healthier on activity? | Yes — deal scores from amount, age, activity, engagement | Activity is mostly seller activity. A rep can send six emails into a dead deal. |
| Has the buyer confirmed they intend to buy? | Only if somebody typed it in | Whether a buyer-side person replied, and what they said |
| Is there a known decision-maker? | A contact record exists, if a rep added one | Whether that person has ever appeared on a thread or a call |
| Is the close date credible? | It reports the date the rep typed | Whether the remaining steps fit in the time left |
Where the CRM is sufficient, and where the record runs out.
Every row in the right-hand column has the same shape. The CRM knows what was recorded about the deal; it does not independently know what the buyer did. That is not a defect — it is what a system of record is for.
The evidence
What a real deal looks like in evidence
Klosure reads the buyer activity already happening around the opportunity and asks a fixed set of questions of it. None of them can be answered by a rep in a pipeline review.
- Has anyone on the buyer’s side replied without being chased, and how long has the thread been one-sided?
- Has the person who signs appeared — on a thread, on a call, on a CC line — or is signing authority still unknown?
- Is there a next meeting in a calendar, or only an intention to reconnect?
- Were the questions the buyer raised ever answered, and were the commitments either side made ever met?
- How long has the deal been in this stage, and did anything the buyer did move it there?
- Is the silence meaningful? Five days quiet in a live negotiation is not five days quiet over Eid.
Each answer carries its evidence. Every factor carries its evidence: a quote of at most 25 words, the source messages and the date — or an explicit absence flag when the missing thing is the signal.
Only buyer signals, system signals, verifiable facts and elapsed time move the score. A rep cannot talk it up by asking for a better number.
The mechanism
Structural gaps cap the number, whoever called the deal
The part that makes a forecast inspectable rather than persuasive is the ceiling. When something structural is missing, confidence cannot exceed a fixed number — not because a model judged it unlikely, but because deterministic code applies the cap and overrides the model.
| Structural gap | Confidence cannot exceed |
|---|---|
| No identified signing authority within 30 days of the deadline | 35% |
| Signing authority unknown at any deadline | 55% |
| Ten days with no buyer-side contact | 45% |
| Fifteen days dark with nothing booked | 25% |
| Three weeks stuck in one stage | 40% |
| A negotiation carrying three or more unresolved blockers | 55% |
Confidence ceilings, exactly as the code applies them. Only the lowest applicable cap applies — they never compound.
Only the lowest applicable cap applies — they never compound. Ceilings are recomputed in deterministic code that overrides the model, then applied again at read time, so silence keeps accruing between reads.
| Category | Capped confidence |
|---|---|
| Commit | 65% and above |
| Best case | 45–64% |
| Pipeline | 25–44% |
Forecast categories are derived from the capped score, not declared by a person.
The category is derived from the capped score rather than declared by anyone. A deal ten days silent is capped at 45% and therefore cannot sit in Commit, whoever called it.
How to check
The test that settles it in twenty minutes
Do not evaluate this on a feature list. Take the deals your managers currently believe will close and ask which ones a system built on buyer evidence would remove from the forecast, and why.
- Pick twenty to thirty live opportunities, deliberately including five or more your team currently calls Commit.
- Include a few that have already slipped once, and a couple that look genuinely healthy.
- Have your managers independently name the five riskiest, without seeing the tool’s answer.
- Compare. The question is not whether the tool sounds clever — it is whether it found risks your managers missed.
- Check again in thirty to sixty days against what actually happened.
If it mostly restates what your managers already knew, you do not need it. That is a real outcome and we would rather you reached it in twenty minutes than after a quarter.
The limits
What this does not do
What this does not do
The score is deliberately not a calibrated statistical probability. The ceilings are rules derived from deal mechanics, not a model trained on your closed history.
Klosure reads email, calendar and meetings. A deal negotiated mainly over phone or WhatsApp shows less signal than one run in the inbox.
Forecasting here is decision support. It does not guarantee an outcome, and it is meant to be argued with on the evidence it shows.
One limit deserves naming here rather than in a footnote: the confidence score is not a statistically calibrated probability trained on your closed-won history. It is a read of the evidence with deterministic ceilings derived from deal mechanics. If you are shopping for "this deal has a 73.4% chance of closing by 30 September", that is a different product and we are not it.
FAQ
Frequently asked questions
Why is my sales forecast a guess?
A forecast is a guess when its inputs are declarations rather than evidence. Stage, probability and close date are values a seller entered, so the forecast is a well-presented summary of what the team believes. Klosure rebuilds the number from buyer signals — replies, silence, stakeholder coverage, unmet commitments — so the forecast can be inspected against what actually happened rather than trusted.
Does HubSpot already tell me which deals are real?
HubSpot tells you which deals look healthier, which is a different thing. Deal scores, forecast categories and weighted pipeline are strong, and Klosure does not replace them. What a CRM cannot do independently is establish that the buyer intends to buy, because it knows what your team recorded rather than what the buyer did. That boundary is where Klosure starts.
Can a rep talk Klosure’s confidence score up?
No. Only buyer signals, system signals, verifiable facts and elapsed time move Klosure’s confidence score. Asking for a better number does nothing, and structural gaps such as unknown signing authority cap the score in deterministic code that overrides the model. A deal ten days silent is capped at 45% and therefore cannot sit in Commit, whoever called it.
Do I need to replace my CRM to fix the forecast?
No, and you should be suspicious of anyone who says otherwise. Klosure runs alongside the CRM rather than replacing it — deals push out one way, leads pull in, and the CRM stays the system of record. If your CRM is not configured properly, fix that first: an intelligence layer on top of an unused pipeline improves nothing.
How long before the forecast changes?
Klosure reads the email, calendar and meeting history already attached to your open deals, so the first read reflects evidence that already exists rather than waiting for new activity to accumulate. What it cannot do is reconstruct a deal that was negotiated somewhere it cannot see — Klosure does not capture phone calls or WhatsApp messages.
Bring one deal. See what Klosure would read.
A 20-minute walkthrough against a real opportunity from your own pipeline — not a canned demo. Bring a deal you are not sure about.